Merge pull request #23221 from WilliamDAssafMSFT/20220714-adr-supportability

20220714 adr pvs supportability

Author: PRMerger19

docs/relational-databases/accelerated-database-recovery-concepts.md

@@ -1,7 +1,7 @@
 ---
 description: "Accelerated database recovery"
 title: "Accelerated database recovery"
-ms.date: "05/24/2022"
+ms.date: 07/14/2022
 ms.prod: sql
 ms.prod_service: backup-restore
 ms.technology: backup-restore
@@ -83,8 +83,8 @@ The ADR recovery process has the same three phases as the current recovery proce
 
 - **Redo** phase
 
-  Broken into two sub-phases
-  - Sub-phase 1
+  Broken into two subphases
+  - Subphase 1
 
       Redo from SLOG (oldest uncommitted transaction up to last checkpoint). Redo is a fast operation as it only needs to process a few records from the SLOG.
 
@@ -156,10 +156,11 @@ There are several improvements to address persistent version store (PVS) storage
 
 - **Multi-threaded version cleanup**  
 
-  In [!INCLUDE[sssql19-md](../includes/sssql19-md.md)], the cleanup process is single threaded within a SQL Server instance. In SQL Server 2022 (16.x) Preview, the cleanup process has multiple threads, with one thread per SQL Server database. 
+  In [!INCLUDE[sssql19-md](../includes/sssql19-md.md)], the cleanup process is single threaded within a SQL Server instance. 
 
-  In [!INCLUDE[sssql22-md](../includes/sssql22-md.md)], CTP 2.0, you can also enable multi-threaded version cleanup at the database level with trace flag 3515. This allows multiple threads for cleanup per database. This improvement is valuable when you have a fewer number of large databases.
-To enable trace flag 3515 for the instance, run the following command:
+  In [!INCLUDE[sssql22-md](../includes/sssql22-md.md)], CTP 2.0, you can also enable multi-threaded version cleanup at the database level with trace flag 3515. This allows multiple threads for cleanup per database. This improvement is valuable when you have multiple large databases.
+
+  To enable trace flag 3515 for the instance, run the following command:
 
    ```sql
    DBCC TRACEON(3515, -1)
@@ -175,6 +176,10 @@ To enable trace flag 3515 for the instance, run the following command:
   RECONFIGURE WITH OVERRIDE; 
   ```
 
+- **New extended event**
+
+  A new extended event, `tx_mtvc2_sweep_stats`, has been added for telemetry on the ADR PVS multi-threaded version cleaner.
+
 ## Best practices and guidance
 
 For guidance on workloads that are and are not recommended for ADR, see [Manage accelerated database recovery](accelerated-database-recovery-management.md).

docs/relational-databases/accelerated-database-recovery-troubleshoot.md

@@ -1,7 +1,7 @@
 ---
 description: "Troubleshoot accelerated database recovery"
 title: "Troubleshoot accelerated database recovery"
-ms.date: 07/12/2022
+ms.date: 07/14/2022
 ms.prod: sql
 ms.prod_service: backup-restore
 ms.technology: backup-restore
@@ -23,7 +23,9 @@ This article helps administrators diagnose issues with accelerated database reco
 
 ## Examine the persistent version store (PVS)
 
-Leverage the `sys.dm_tran_persistent_version_store_stats` DMV to identify if the size of the accelerated database recovery (ADR) PVS is growing larger than expected, and then to determine which factor is preventing persistent version store (PVS) cleanup.
+Leverage the [sys.dm_tran_persistent_version_store_stats](system-dynamic-management-views/sys-dm-tran-persistent-version-store-stats.md) DMV to identify if the size of the accelerated database recovery (ADR) PVS is growing larger than expected, and then to determine which factor is preventing persistent version store (PVS) cleanup.
+
+Included in the following sample script is the column `sys.dm_tran_persistent_version_store_stats.pvs_off_row_page_skipped_oldest_aborted_xdesid`, which was added in [!INCLUDE[sssql22-md](../includes/sssql22-md.md)] and contains the number of pages skipped for reclaim due to oldest aborted transactions. If the version cleaner is slow or invalidated, this will reflect how many pages must be kept for aborted transactions.
 
 The sample query shows all information about the cleanup processes and shows the current PVS size, oldest aborted transaction, and other details:
 
@@ -41,7 +43,8 @@ SELECT
  asdt.session_id AS active_transaction_session_id,
  asdt.elapsed_time_seconds AS active_transaction_elapsed_time_seconds,
  pvss.pvs_off_row_page_skipped_low_water_mark,
- pvss.pvs_off_row_page_skipped_min_useful_xts
+ pvss.pvs_off_row_page_skipped_min_useful_xts,
+ pvss.pvs_off_row_page_skipped_oldest_aborted_xdesid -- SQL Server 2022 only
 FROM sys.dm_tran_persistent_version_store_stats AS pvss
 CROSS APPLY (SELECT SUM(size*8.) AS total_db_size_kb FROM sys.database_files WHERE [state] = 0 and [type] = 0 ) AS df 
 LEFT JOIN sys.dm_tran_database_transactions AS dt
@@ -148,13 +151,19 @@ For example,
 EXEC sys.sp_persistent_version_cleanup [WideWorldImporters];
 ```
 
+## Use trace flags to capture cleanup failures
+ 
+  *Applies to [!INCLUDE[sql-server-2022](../includes/sssql22-md.md)] and later*
+
+  Trace flag 4025 can be enabled to record ADR PVS cleanup behavior in the SQL Server error log. Typically this would result in a new log event recorded every 10 minutes. For more information, see [DBCC TRACEON (Transact-SQL)](../t-sql/database-console-commands/dbcc-traceon-transact-sql.md).
+
 ## See also
 
 - [sys.sp_persistent_version_cleanup](system-stored-procedures/sys-sp-persistent-version-cleanup-transact-sql.md)
 - [sys.dm_tran_persistent_version_store_stats](system-dynamic-management-views/sys-dm-tran-persistent-version-store-stats.md)
 - [sys.dm_tran_aborted_transactions](system-dynamic-management-views/sys-dm-tran-aborted-transactions.md)
 
-## Next steps 
+## Next steps
 
 - [Accelerated database recovery concepts](accelerated-database-recovery-concepts.md)
 - [Manage accelerated database recovery](accelerated-database-recovery-management.md)